bartowski commited on
Commit
2301fcd
1 Parent(s): 6ce3b5e

measurement.json

Browse files
Files changed (2) hide show
  1. README.md +71 -0
  2. measurement.json +0 -0
README.md ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: apache-2.0
4
+ datasets:
5
+ - Locutusque/Hercules-v3.0
6
+ tags:
7
+ - chemistry
8
+ - biology
9
+ - code
10
+ - medical
11
+ - not-for-all-audiences
12
+ quantized_by: bartowski
13
+ pipeline_tag: text-generation
14
+ ---
15
+
16
+ ## Exllama v2 Quantizations of Hercules-3.1-Mistral-7B
17
+
18
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.13">turboderp's ExLlamaV2 v0.0.13</a> for quantization.
19
+
20
+ <b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
21
+
22
+ Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
23
+
24
+ Original model: https://huggingface.co/Locutusque/Hercules-3.1-Mistral-7B
25
+
26
+ | Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
27
+ | ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
28
+ | [8_0](https://huggingface.co/bartowski/Hercules-3.1-Mistral-7B-exl2/tree/8_0) | 8.0 | 8.0 | 8.4 GB | 9.8 GB | 11.8 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
29
+ | [6_5](https://huggingface.co/bartowski/Hercules-3.1-Mistral-7B-exl2/tree/6_5) | 6.5 | 8.0 | 7.2 GB | 8.6 GB | 10.6 GB | Very similar to 8.0, good tradeoff of size vs performance, **recommended**. |
30
+ | [5_0](https://huggingface.co/bartowski/Hercules-3.1-Mistral-7B-exl2/tree/5_0) | 5.0 | 6.0 | 6.0 GB | 7.4 GB | 9.4 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
31
+ | [4_25](https://huggingface.co/bartowski/Hercules-3.1-Mistral-7B-exl2/tree/4_25) | 4.25 | 6.0 | 5.3 GB | 6.7 GB | 8.7 GB | GPTQ equivalent bits per weight, slightly higher quality. |
32
+ | [3_5](https://huggingface.co/bartowski/Hercules-3.1-Mistral-7B-exl2/tree/3_5) | 3.5 | 6.0 | 4.7 GB | 6.1 GB | 8.1 GB | Lower quality, only use if you have to. |
33
+
34
+ ## Download instructions
35
+
36
+ With git:
37
+
38
+ ```shell
39
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/Hercules-3.1-Mistral-7B-exl2 Hercules-3.1-Mistral-7B-exl2-6_5
40
+ ```
41
+
42
+ With huggingface hub (credit to TheBloke for instructions):
43
+
44
+ ```shell
45
+ pip3 install huggingface-hub
46
+ ```
47
+
48
+ To download the `main` (only useful if you only care about measurement.json) branch to a folder called `Hercules-3.1-Mistral-7B-exl2`:
49
+
50
+ ```shell
51
+ mkdir Hercules-3.1-Mistral-7B-exl2
52
+ huggingface-cli download bartowski/Hercules-3.1-Mistral-7B-exl2 --local-dir Hercules-3.1-Mistral-7B-exl2 --local-dir-use-symlinks False
53
+ ```
54
+
55
+ To download from a different branch, add the `--revision` parameter:
56
+
57
+ Linux:
58
+
59
+ ```shell
60
+ mkdir Hercules-3.1-Mistral-7B-exl2-6_5
61
+ huggingface-cli download bartowski/Hercules-3.1-Mistral-7B-exl2 --revision 6_5 --local-dir Hercules-3.1-Mistral-7B-exl2-6_5 --local-dir-use-symlinks False
62
+ ```
63
+
64
+ Windows (which apparently doesn't like _ in folders sometimes?):
65
+
66
+ ```shell
67
+ mkdir Hercules-3.1-Mistral-7B-exl2-6.5
68
+ huggingface-cli download bartowski/Hercules-3.1-Mistral-7B-exl2 --revision 6_5 --local-dir Hercules-3.1-Mistral-7B-exl2-6.5 --local-dir-use-symlinks False
69
+ ```
70
+
71
+ Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski
measurement.json ADDED
The diff for this file is too large to render. See raw diff